An FPGA-Based Convolutional Neural Network Coprocessor

Table 3

The computational performance comparison results for different platforms.

Layer group	This work (GOP/s) VGG16	[15]	[16]	[17]
Conv_1	23.04	46.42	1578.8	45.15
Conv_2	52.97	80.44	1675.5	69.60
Conv_3	88.45	151.57	2177.1	103.51
Conv_4	111.47	147.91	2791.6	86.04
Conv_5	66.70	186.99	1003.5	36.27